用apache + wsgi + jdango 部署调用PaddleOCR的程序，执行到代码“from paddleocr import PaddleOCR”时报错

[accsoar]

• 系统环境/System Environment：windows server 2019 64bit
• 版本号/Version：
  • Paddle: 2.6.0 (installed by C:\Python310>python -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple)
  • PaddleOCR：2.7.0.3 (installed by pip3 install "paddleocr>=2.0.1")
• 问题相关组件/Related components：
  • Apache: 2.4.58 amd64
  • Python: 3.10
  • mode_wsgi: mod_wsgi-4.9.2-cp310-cp310-win_amd64.whl
  • django: 5.0.3 (installed by pip3 install django)
• 运行指令/Command Code：
• 完整报错/Complete Error Message： Apache报错如下：
  [Fri Mar 15 11:55:55.282321 2024] [wsgi:error] [pid 3872:tid 1208] [client 10.235.59.122:59096] Error: Can not import paddle core while this file exists: C:\Python310\lib\site-packages\paddle\base\libpaddle.pyd\r

用apache + wsgi + jdango部署的web service，提供一个API来使用PaddleOCR识别图像

API涉及的代码如下（ocr识别图片的代码已移除，仅保留这句发生问题的代码“from paddleocr import PaddleOCR”）：

from django.http import HttpRequest
from django.http import HttpResponse
from django.views.decorators.csrf import csrf_exempt

import time

from paddleocr import PaddleOCR

@csrf_exempt
def recognize(request):
    print("ping() is called")

    return HttpResponse("{}",
                        content_type="application/json; encoding=utf-8",
                        status=200)

1. 此django程序用python本地启用后（python manage.py runserver），API调用成功（ocr识别图片代码不删除的话，也可成功识别图像）。

2. 但部署到Apache后，调用API时apache报错，报错log如下：

[Fri Mar 15 11:54:47.666727 2024] [core:notice] [pid 2116:tid 464] AH00094: Command line: 'C:\\Apache24\\bin\\httpd.exe -d C:/Apache24'
[Fri Mar 15 11:54:47.666727 2024] [mpm_winnt:notice] [pid 2116:tid 464] AH00418: Parent: Created child process 3872
[Fri Mar 15 11:54:49.104244 2024] [mpm_winnt:notice] [pid 3872:tid 484] AH00354: Child: Starting 64 worker threads.
[Fri Mar 15 11:55:55.282321 2024] [wsgi:error] [pid 3872:tid 1208] [client 10.235.59.122:59096] Error: Can not import paddle core while this file exists: C:\\Python310\\lib\\site-packages\\paddle\\base\\libpaddle.pyd\r
[Fri Mar 15 11:55:56.391697 2024] [wsgi:error] [pid 3872:tid 1208] [client 10.235.59.122:59096] Error: Can not import paddle core while this file exists: C:\\Python310\\lib\\site-packages\\paddle\\base\\libpaddle.pyd\r
[Fri Mar 15 11:55:56.860444 2024] [mpm_winnt:notice] [pid 2116:tid 464] AH00428: Parent: child process 3872 exited with status 255 -- Restarting.

但把那句from paddleocr import PaddleOCR删除的话，API调用也是成功的 (也就是说，我的Django程序在Apache中的wsgi配置应该是没问题的)

在Apache的httpd.conf文件中，wsgi相关配置如下（相关路径配置确认都没问题）：

Define SRVROOT "c:/Apache24"

Listen 28886

LoadModule wsgi_module "C:/Python310/lib/site-packages/mod_wsgi/server/mod_wsgi.cp310-win_amd64.pyd"

WSGIPythonHome "C:/Python310"

#指定website的wsgi.py配置文件路径  
WSGIScriptAlias / C:/xxp/dj_paddleocr_test/dj_paddleocr_test/wsgi.py

#指定项目路径  
WSGIPythonPath  C:/xxp/dj_paddleocr_test

<Directory C:/xxp/dj_paddleocr_test/dj_paddleocr_test>  
    WSGIPassAuthorization On
    Options None
    AllowOverride None
    <Files wsgi.py>  
        Require all granted  
    </Files>

    # Limit HTTP methods
    <LimitExcept GET POST OPTIONS>
        Require all denied
    </LimitExcept>
</Directory>  

[changdazhou]

建议尝试一下使用源码安装paddleocr哈

[accsoar]

建议尝试一下使用源码安装paddleocr哈

就是说，直接下载paddleocr的源码？那请问如何在我的django程序中调用paddleocr来识别图像呢？有类似的示例程序吗？谢谢

[changdazhou]

对的，直接下载源码，使用最新的源码编译成whl包安装后再进行尝试哈